Exploring Sources of Uncertainties in Solar Resource Measurements

1. Exploring Sources of Uncertainties in
Solar Resource Measurements
Presenter: Manajit Sengupta
Authors: Aron Habte, Manajit Sengupta
2016 6th PV Performance and Monitoring Workshop
Freiburg, Germany (October 24-25, 2016)

2. 2
The right observations of
wind and solar resources
Measurements
Targeted predictions of
resources and
plant performance
Modeling
Raising everyone to
the same level and
enabling dialog
Standards
Provide high-quality meteorological and power data for energy
yield assessment, resource characterization, and grid integration
Sensing, Measurement, and Forecasting

3. 3
• NREL’s Sensing, Measurement, and Forecasting Group collects and
disseminates accurate solar resource measurements.
• Best practices for solar resources measurement, calibration, and
characterization are followed.
• Advancing best practices benefits solar conversion projects by
improving the bankability of the underlying data.
• The accuracy of solar resource measurements depends on:
o Instrument specifications
o Calibration procedures
o Measurement setup
o Maintenance (cleaning)
o Location and environmental conditions.
Why Explore Sources of Uncertainty?

4. 4
3. Standard
uncertainty
4.
Sensitivity
Coefficient
5.
Combined
Uncertainty
6.
Expanded
Uncertainty
1.
Measurement
Equation
2. Sources of
Uncertainties
Measurement
Uncertainty
Estimation
Sources of Measurement
Uncertainty
• Calibration
• Spectral response
• Zenith angle response
• Maintenance----Soiling
• Data logger uncertainty
• Temperature dependence
• Nonlinear response
• Thermal offset
• Instrument aging
• Understanding and quantifying each source of uncertainty is essential for the
determination of overall uncertainty.
Measurement Uncertainty Estimation

5. Evaluating Calibration Methods

6. 6
• Both indoor and outdoor methods are traceable to the
World Radiometric Reference.
• Indoor calibration of radiometers provides:
o Control of test conditions
o Calibration results independent of outdoor conditions
o Convenience.
• Outdoor calibrations are useful for cosine response
correction, which ultimately assists in reducing
measurement uncertainty.
Overview

7. 7
a RESPONSIVITY VALUE CASES APPLIED IN THE STUDY. WHEN THERMAL OFFSET CORRECTION IS APPLICABLE (YES),
EQUATION (3) IS USED. IF NOT APPLICABLE (NO), EQUATION (4) IS USED.
Cases Calibration Method
Thermal Offset Correction Applicability
Thermopile
Pyranometer
Thermopile
Pyrheliometer
Silicon Photodiode
Pyranometer
Case 1
BORCALb responsivity as a function
of solar zenith angle (SZA) Yes No No
Case 2
Manufacturer calibration
responsivity at manufacturer-
specified SZA in degrees N/A N/A N/A
Case 3 BORCAL responsivity at 45° Yes No No
Case 4 BORCAL responsivity at 45° No No No
Case 5
Manufacturer calibration
responsivity at manufacturer-
specified SZA in degrees with
manufacturer-supplied
measurement equation N/A N/A N/A
a The study is published in Progress in Photovoltaics:
http://onlinelibrary.wiley.com/doi/10.1002/pip.2812/full.
b Broadband Outdoor Radiometer Calibrations
Ten months of 1-minute data for clear-sky conditions
(KN>0.6) from 12 radiometers were compared.
Calibration Methods

8. 8
 CMP22 has relatively small difference among all the methods compared to the MS-410
and SPP radiometers.
 For photodiode pyranometers, the manufacturer-supplied responsivities have higher
deviation.
Each colored box shows the interquartile range and represents a 10-degree zenith bin. The circle in each blue box
represents the mean, and the black horizontal line represents the median value. Ninety-nine percent of the data lies
within the whiskers. Data beyond the whiskers are plotted with dots.
(1) BORCAL: Function of SZA, (2) manufacturer-specified SZA in degrees, (3) BORCAL responsivity at 45° with thermal offset
correction, (4) BORCAL responsivity at 45° without thermal offset correction, (5) manufacturer-specified SZA in degrees
with manufacturer-supplied measurement equation.
Thermopile Pyranometers Photodiode Pyranometers
GHI: Measurement Differences from Calibration

9. 9
 The sNIP pyrheliometer data show better agreement to the reference direct normal irradiance (DNI)
(CHP1) data than the DR02 and MS-56 pyrheliometers.
 The NREL responsivity function method provides better results for the DR02 radiometer than the
factory responsivity method.
(1) BORCAL: Function of SZA, (2) manufacturer-specified SZA in degrees, (3) BORCAL responsivity at 45° with thermal
offset correction, (4) BORCAL responsivity at 45° without thermal offset correction, (5) manufacturer-specified SZA
in degrees with manufacturer-supplied measurement equation.
Thermopile Pyrheliometers
DNI: Measurement Differences from Calibration

10. Quantifying Spectral Error

11. 11
• In the International Standards Organization (ISO) and
World Meteorological Organization (WMO) “spectral
selectivity” term is the only specification that does
not translate directly into a measurement error.
• This is a problem in uncertainty evaluation.
Overview

12. 12
Spectral Mismatch Equation
• τdome= Dome transmittance
• 𝜶𝜶(𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑐𝑐 𝑐𝑐)= Absorptance of coating
• EAMi
= Spectral irradiance under various air mass
(obtained using SMARTS)
• EAM1.41
= Reference spectral data at AM 1.41 (SZA 45).

13. 13
5. TSP-1
1. PSP 2. PSP
3. PSP
4. PSP
6. Hukseflux
(Data from manufacturer)
7&8. Kipp &
Zonen
(Data from manufacturer)
9. N-WG295
Inst# Model Type Comment
1 PSP Double dome and aged coating
2 PSP Double dome and aged coating
3 PSP Double dome and aged coating
4 PSP Double dome and aged coating
5 TSP-1 Double dome and aged coating
6 --- Transmission 2 mm and new coating data (Hukseflux) Provided by manufacturer
7 --- Transmission 4 mm (Kipp & Zonen) Provided by manufacturer
8 --- Transmission 4 mm + Fresnel (Kipp & Zonen) Provided by manufacturer
9 --- SCHOTT-N-WG295 Data sheet
Radiometers Included in the Study

14. 14
DATA SHEET, NEWMEASURED, AGED
In-door Domes Transmittance Measurement
Indoor Coating Absorptance
Measurement:
Indoor measurement
ASD—reflectance
Labsphere reference plaque
Flat black cloth
Transmittance and Absorptance Measurement

15. 15
• Results are based on combined
transmittance measurement of the inner and
outer dome for Inst# 1–5.
• Numbers 1–9 are instrument numbers and
10 locations under different air mass.
• Numbers 6–9 are new radiometers with new
glass transmittance and coating absorptance.
—Data obtained from the manufacturers.
Result Using Indoor Transmittance Measurement (400–2,400 nm)

16. Quantifying Soiling Effects

17. 17
• Artificial soiling that simulates various environments complements and/or
substitutes natural soiling determination.
• Various degrees of soiling reduce the optical transmittance of the glass
dome of the pyranometer, which ultimately reduces the detector output
(energy loss).
• The study demonstrates how cleaning radiometers is essential in obtaining
accurate radiometric data.
• The study is beneficial for overall measurement uncertainty estimation of
radiometric data.
• The study will also assist meteorological station operators in estimating
the irradiance reduction due to soiling by comparing the images of the
artificial soiling to the field conditions.
Overview

18. 18
S/N: 29353F3
Simulated Snow/Dew
Fourteen artificially soiled pyranometer domes were measured.
Artificial Soiling: Various Types and Levels of Soiling

19. 19
 ASD spectroradiometer was used to measure the
transmittance (350–2,400 nm).
 Stable light source was used to measure the transmittance.
 Twelve-inch integrating sphere was used.
Light out
Stable light source
• Working toward the development of a standardized artificial soiling method for
thermopile radiometers:
Twelve-inch integrating
sphere
Light in
Fiber-optic going
through the dome
holder
Source: http://www.nrel.gov/docs/fy16osti/66792.pdf
Method: Indoor Measurement

20. 20
Result

21. 21
• Solar resource data with known and traceable uncertainty estimates
are essential for the site selection of renewable energy technology
deployment, system design, system performance, and system
operations.
• Developing consensus methodologies for determining solar resource
measurement uncertainties is essential in obtaining accurate
radiometric data.
• Calibration differences between manufacturers’ and outdoor NREL
BORCAL provided irradiance differences up to 1%–2% for
pyranometers and less than 1% for pyrheliometers.
• Spectral mismatch contributes to spectral error up to 1.6% for indoor
transmittance measurement.
• Various degrees of soiling reduce the optical transmittance of the
glass dome of the pyranometer, which ultimately reduces the
detector output (energy loss). The observed reduction was 0.2%–
27%.
Summary

22. Questions?
Thank you!
manajit@nrel.gov
Sensing, Measurement, and Forecasting Group
Power Systems Engineering Center
National Renewable Energy Laboratory
Office: 303-275-3706 | Fax:303-275-3835
Note: Except as otherwise indicated, all images are NREL owned.